We do the same thing with Redshift. Using temp tables also gives us more direct control over the query plan, since we can specify distribution and sort keys. On

We do the same thing with Redshift. Using temp tables also gives us more direct control over the query plan, since we can specify distribution and sort keys. On top of that, we can run unit tests on the temporary table (explained in another comment) and fail the query if tests fail.